Implement webhook validations for the PyTorchJob #2035
+748
−348
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -19,6 +19,7 @@ package main | |
| import ( | ||
| "errors" | ||
| "flag" | ||
| "net/http" | ||
| "os" | ||
| "strings" | ||
|
|
||
|
|
@@ -40,9 +41,11 @@ import ( | |
| volcanoclient "volcano.sh/apis/pkg/client/clientset/versioned" | ||
|
|
||
| kubeflowv1 "github.com/kubeflow/training-operator/pkg/apis/kubeflow.org/v1" | ||
| "github.com/kubeflow/training-operator/pkg/cert" | ||
| "github.com/kubeflow/training-operator/pkg/config" | ||
| controllerv1 "github.com/kubeflow/training-operator/pkg/controller.v1" | ||
| "github.com/kubeflow/training-operator/pkg/controller.v1/common" | ||
| "github.com/kubeflow/training-operator/pkg/webhooks" | ||
| //+kubebuilder:scaffold:imports | ||
| ) | ||
|
|
||
|
|
@@ -72,8 +75,11 @@ func main() { | |
| var enabledSchemes controllerv1.EnabledSchemes | ||
| var gangSchedulerName string | ||
| var namespace string | ||
| var controllerThreads int | ||
| var webhookServerPort int | ||
| var webhookServiceName string | ||
| var webhookSecretName string | ||
|
|
||
| flag.StringVar(&metricsAddr, "metrics-bind-address", ":8080", "The address the metric endpoint binds to.") | ||
| flag.StringVar(&probeAddr, "health-probe-bind-address", ":8081", "The address the probe endpoint binds to.") | ||
| flag.BoolVar(&enableLeaderElection, "leader-elect", false, | ||
|
|
@@ -86,7 +92,6 @@ func main() { | |
| " Note: If you set another scheduler name, the training-operator assumes it's the scheduler-plugins.") | ||
| flag.StringVar(&namespace, "namespace", os.Getenv(EnvKubeflowNamespace), "The namespace to monitor kubeflow jobs. If unset, it monitors all namespaces cluster-wide."+ | ||
| "If set, it only monitors kubeflow jobs in the given namespace.") | ||
| flag.IntVar(&controllerThreads, "controller-threads", 1, "Number of worker threads used by the controller.") | ||
|
|
||
| // PyTorch related flags | ||
|
|
@@ -101,6 +106,11 @@ func main() { | |
| flag.StringVar(&config.Config.MPIKubectlDeliveryImage, "mpi-kubectl-delivery-image", | ||
| config.MPIKubectlDeliveryImageDefault, "The image for mpi launcher init container") | ||
|
|
||
| // Cert generation flags | ||
| flag.IntVar(&webhookServerPort, "webhook-server-port", 9443, "Endpoint port for the webhook server.") | ||
| flag.StringVar(&webhookServiceName, "webhook-service-name", "training-operator", "Name of the Service used as part of the DNSName") | ||
| flag.StringVar(&webhookSecretName, "webhook-secret-name", "training-operator-webhook-cert", "Name of the Secret to store CA and server certs") | ||
|
|
||
| opts := zap.Options{ | ||
| Development: true, | ||
| StacktraceLevel: zapcore.DPanicLevel, | ||
|
|
@@ -124,9 +134,9 @@ func main() { | |
| Metrics: metricsserver.Options{ | ||
| BindAddress: metricsAddr, | ||
| }, | ||
| WebhookServer: webhook.NewServer(webhook.Options{ | ||
| Port: webhookServerPort, | ||
| }), | ||
| HealthProbeBindAddress: probeAddr, | ||
| LeaderElection: enableLeaderElection, | ||
| LeaderElectionID: leaderElectionID, | ||
|
|
@@ -137,30 +147,36 @@ func main() { | |
| os.Exit(1) | ||
| } | ||
|
|
||
| certsReady := make(chan struct{}) | ||
| defer close(certsReady) | ||
| certGenerationConfig := cert.Config{ | ||
| WebhookSecretName: webhookSecretName, | ||
| WebhookServiceName: webhookServiceName, | ||
| } | ||
| if err = cert.ManageCerts(mgr, certGenerationConfig, certsReady); err != nil { | ||
| setupLog.Error(err, "Unable to set up cert rotation") | ||
| os.Exit(1) | ||
| } | ||
|
|
||
| setupProbeEndpoints(mgr, certsReady) | ||
| // Set up controllers using goroutines to start the manager quickly. | ||
| go setupControllers(mgr, enabledSchemes, gangSchedulerName, controllerThreads, certsReady) | ||
|
|
||
| //+kubebuilder:scaffold:builder | ||
|
|
||
| setupLog.Info("starting manager") | ||
| if err := mgr.Start(ctrl.SetupSignalHandler()); err != nil { | ||
| setupLog.Error(err, "problem running manager") | ||
| os.Exit(1) | ||
| } | ||
| } | ||
|
|
||
| func setupControllers(mgr ctrl.Manager, enabledSchemes controllerv1.EnabledSchemes, gangSchedulerName string, controllerThreads int, certsReady <-chan struct{}) { | ||
| setupLog.Info("Waiting for certificate generation to complete") | ||
| <-certsReady | ||
| setupLog.Info("Certs ready") | ||
|
|
||
| setupLog.Info("registering controllers...") | ||
| // Prepare GangSchedulingSetupFunc | ||
| gangSchedulingSetupFunc := common.GenNonGangSchedulerSetupFunc() | ||
| if strings.EqualFold(gangSchedulerName, string(common.GangSchedulerVolcano)) { | ||
|
|
@@ -182,15 +198,52 @@ func setupControllers(mgr ctrl.Manager, enabledSchemes controllerv1.EnabledSchem | |
| } | ||
| errMsg := "failed to set up controllers" | ||
| for _, s := range enabledSchemes { | ||
| setupReconcilerFunc, supportedReconciler := controllerv1.SupportedSchemeReconciler[s] | ||
| if !supportedReconciler { | ||
| setupLog.Error(errors.New(errMsg), "scheme is not supported", "scheme", s) | ||
| os.Exit(1) | ||
| } | ||
| if err := setupReconcilerFunc(mgr, gangSchedulingSetupFunc, controllerThreads); err != nil { | ||
| setupLog.Error(errors.New(errMsg), "unable to create controller", "scheme", s) | ||
| os.Exit(1) | ||
| } | ||
| setupWebhookFunc, supportedWebhook := webhooks.SupportedSchemeWebhook[s] | ||
| if !supportedWebhook { | ||
| setupLog.Error(errors.New(errMsg), "scheme is not supported", "scheme", s) | ||
| os.Exit(1) | ||
| } | ||
| if err := setupWebhookFunc(mgr); err != nil { | ||
| setupLog.Error(errors.New(errMsg), "unable to start webhook server", "scheme", s) | ||
| os.Exit(1) | ||
| } | ||
| } | ||
| } | ||
|
|
||
| func setupProbeEndpoints(mgr ctrl.Manager, certsReady <-chan struct{}) { | ||
| defer setupLog.Info("Probe endpoints are configured on healthz and readyz") | ||
|
|
||
| if err := mgr.AddHealthzCheck("healthz", healthz.Ping); err != nil { | ||
| setupLog.Error(err, "unable to set up health check") | ||
| os.Exit(1) | ||
| } | ||
|
|
||
| // Wait for the webhook server to be listening before advertising the | ||
| // training-operator replica as ready. This allows users to wait with sending the first | ||
| // requests, requiring webhooks, until the training-operator deployment is available, so | ||
| // that the early requests are not rejected during the traininig-operator's startup. | ||
| // We wrap the call to GetWebhookServer in a closure to delay calling | ||
| // the function, otherwise a not fully-initialized webhook server (without | ||
| // ready certs) fails the start of the manager. | ||
| if err := mgr.AddReadyzCheck("readyz", func(req *http.Request) error { | ||
| select { | ||
| case <-certsReady: | ||
| return mgr.GetWebhookServer().StartedChecker()(req) | ||
|
There was a problem hiding this comment. Can you explain about this closure? how is There was a problem hiding this comment. @johnugeorge If we don't have this extended readiness probe, the controller-runtime readiness server responds "Ready" even if the training operator is still preparing due to certificate generation. Also, you can find the details here: https://github.com/kubernetes-sigs/kueue/pull/1676/files#r1472752662 |
||
| default: | ||
| return errors.New("certificates are not ready") | ||
| } | ||
| }); err != nil { | ||
| setupLog.Error(err, "unable to set up ready check") | ||
| os.Exit(1) | ||
| } | ||
| } | ||
|
|
||
|
|
||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,15 @@ | ||
| apiVersion: kustomize.config.k8s.io/v1beta1 | ||
| kind: Kustomization | ||
| resources: | ||
| - manifests.yaml | ||
| commonLabels: | ||
| control-plane: kubeflow-training-operator | ||
andreyvelich marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| patches: | ||
| - path: patch.yaml | ||
| target: | ||
| group: admissionregistration.k8s.io | ||
| version: v1 | ||
| kind: ValidatingWebhookConfiguration | ||
|
|
||
| configurations: | ||
| - kustomizeconfig.yaml | ||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,10 @@ | ||
| # the following config is for teaching kustomize where to look at when substituting vars. | ||
| # It requires kustomize v2.1.0 or newer to work properly. | ||
| namespace: | ||
| - kind: ValidatingWebhookConfiguration | ||
| group: admissionregistration.k8s.io | ||
| path: webhooks/clientConfig/service/namespace | ||
| create: true | ||
|
|
||
| varReference: | ||
| - path: metadata/annotations |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,26 @@ | ||
| --- | ||
|
There was a problem hiding this comment. Can we name this file as Or kubebuilder sets this name by default ? There was a problem hiding this comment.
Yea, this name is default value 😓 |
||
| apiVersion: admissionregistration.k8s.io/v1 | ||
| kind: ValidatingWebhookConfiguration | ||
| metadata: | ||
| name: validating-webhook-configuration | ||
andreyvelich marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| webhooks: | ||
| - admissionReviewVersions: | ||
| - v1 | ||
| clientConfig: | ||
| service: | ||
| name: webhook-service | ||
| namespace: system | ||
|
There was a problem hiding this comment. Make it There was a problem hiding this comment. We have no way to modify this namespace since this manifest is automatically generated by the controller-gen. But, this namespace is replaced by https://github.com/kubeflow/training-operator/blob/a618d1c3f77f769da7d4f535e25db88d64aadb88/manifests/base/webhook/kustomizeconfig.yaml. |
||
| path: /validate-kubeflow-org-v1-pytorchjob | ||
|
There was a problem hiding this comment. Do we need the full path? Maybe just: There was a problem hiding this comment. We need to follow this form based on the https://book.kubebuilder.io/reference/markers/webhook. |
||
| failurePolicy: Fail | ||
andreyvelich marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| name: validator.pytorchjob.training-operator.kubeflow.org | ||
| rules: | ||
| - apiGroups: | ||
| - kubeflow.org | ||
| apiVersions: | ||
| - v1 | ||
| operations: | ||
| - CREATE | ||
| - UPDATE | ||
| resources: | ||
| - pytorchjobs | ||
| sideEffects: None | ||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,6 @@ | ||
| - op: replace | ||
| path: /webhooks/0/clientConfig/service/name | ||
| value: training-operator | ||
| - op: replace | ||
| path: /metadata/name | ||
| value: validator.training-operator.kubeflow.org |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do you need to check if certs are ready in
setupControllersfunc ?Maybe we can check it before we run the func and not pass
certsReadyto thesetupControllersfunction:training-operator/cmd/training-operator.v1/main.go
Line 162 in a618d1c
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We need this checking since the
setupControlleris started with the go routine.By doing this, we can improve the start-up time and avoid readiness probe errors (#1841) since the manager can start unless not waiting for the completion of setting up controllers.